RTX 4060M and RTX 4070M Are On Par With RTX 3060 and RTX 3070 Desktop Respectively
Yesterday, NVIDIA unveiled its RTX 4070 Ti along with the Ada Lovelace Mobile lineup. Surprisingly, NVIDIA decided to showcase all mobile GPUs in a single go. Some major performance claims were made yesterday, but the data is just not enough to consider them as a reference. In any case, the initial benchmarks of the RTX 4090M and RTX 4080M showed a very close resemblance to the RTX 3090 and RTX 3080 desktop. Today, the RTX 4060M and RTX 4070M have been tested in Geekbench.
RTX 4060M and RTX 4070M
To get some context, we highly suggest you read our detailed coverage regarding these GPUs here. But to keep things short, let’s go over the specifications of both the RTX 4060M and the RTX 4070M.
The RTX 4070M is based on the AD106 GPU, which is expected to be featured on the RTX 4060 desktop. It packs 4608 Cuda cores and can deliver around 20 TFLOPs of FP32 compute performance. This GPU has 12GB of GDDR6 memory and features a TDP ranging from 35W–115W depending on the laptop.
The RTX 4060M on the other hand is equipped with NVIDIA’s AD107 GPU and features 3072 Cuda cores. Delivering around 14.6 TFLOPs of FP32 compute performance, it is actually faster than the RTX 3060 desktop in this segment. The memory has been decreased to 8GB when compared to the 4070M and the wattage will vary from 35W–115W.
Performance
The same laptop is used for both of these tests. The device in question is a next-gen offering from the Asian brand, ‘ThundeRobot‘. Yes, that’s thunder without the ‘r’. Both laptops use Intel’s upcoming i9-13900HX, which we covered yesterday. The RTX 4060M-based laptop used 16GB of memory, whereas the RTX 4070M-equipped laptop featured 32GB of memory.
Now coming, to the performance, the RTX 4060M manages to score 101850 points in Geekbench using the OpenCL API. The RTX 3060 desktop scores around 95,000 points in the same test. So the RTX 4060M is almost 6-7% faster than the RTX 3060, at least in Geekbench. In fact, this score comes quite close to the RTX 2080 from Turing.
The RTX 4070M on the other hand, is 12% faster than the RTX 4060M, as it scores 114074 points. That is quite slower than the RTX 3070 desktop, in fact, it barely manages to reach the RTX 3060 Ti. Making matters worse, the RTX 4080M turns out to be more than 50% faster than the RTX 4070M. However, this leak does not confirm the power consumption of the 4070M used. Therefore, this benchmark will not represent the actual performance.
From our own observation, we can conclude the following:
- RTX 4090M = RTX 3090 Desktop
- RTX 4080M = RTX 3080 Desktop
- RTX 4070M <= RTX 3070 Desktop
- RTX 4060M >= RTX 3060 Desktop
Once again, the TDP will heavily influence the performance of the abovementioned GPUs. Desktop Graphics Cards have a fixed TGP or TDP, which helps compare GPUs easily. However, laptop or mobile GPUs have varying TDPs depending on the laptop you purchase. Once these laptops actually go live, we can compare their performance relative to the power consumption for a more accurate comparison.
